pose a two-stage genome-wide association study of colorectal cancer using the established resources of the NCI-supported Cooperative Family Registry for Colorectal Cancer Studies (Colon CFR). Specifically, we propose the following Aims. Aim 1: Perform a genome-wide association study for CRC using a population-based case-control study design. This aim will be accomplished by genotyping 992 population-based cases and over 1,020 age - and family history- stratified population-based controls recruited through the Colon CFR with >500,000 SNPs spaced on average 1 every 5 kb in the genome. Only microsatellite stable (MSS) or microsatellite instability low (MSI-L) phenotype and mismatch repair (MMR) mutation negative cases will be included. The primary analysis includes single-SNP association tests and haplotype-based analyses that represent associations with all ~4 million common variants found in the human genome. SNPs and haplotypes will be rank-ordered using a novel selection strategy based on statistical significance and the incorporation of prior genomic data into a hierarchical modeling procedure to yield a subset of SNPs that is more likely to represent "true positive" and "important" associations. Aim 2: Positive associations will be confirmed in the same case-control CRC population using additional SNPs. Using the information generated through Aim 1, we will confirm any association between SNPs in the 1,000 most significant regions and CRC risk in this same study sample using on average 4 additional SNPs from each region. Selection of SNPs (either htSNPs or tSNPs), will be based on information from the International HapMap Project and other similar projects (e.g., Perlegen), as well as selecting variants with known or predicted functional effects (e.g., previously implicated in CRC or non-synonymous SNPs, respectively). The additional SNPs will be genotyped and associations determined using all cases and controls analyzed in Aim 1. Aim 3: Positive associations will be validated in a second independent family-based case-control CRC population. Using information from Aims 1 and 2, the most informative SNPs will be identified for each region and association validated in 612 additional familial MSS or MSI-L cases and 950 same-generation relative controls (unaffected siblings, half-siblings, and cousins). Association will be determined using all cases and controls analyzed in Aims 1, 2 and 3. To identify variants that are most strongly associated with CRC risk, single-variant analyses and haplotype-specific tests will be performed jointly considering evidence of association from both stages, and interactions with environmental exposures. Aim 4: Positive associations will be validated among all available pedigree members in families from Aim 3. A final analysis will incorporate additional genotyping on all available pedigree members (~3,000 total subjects, including cases and controls) for the subset of genomic regions/SNPs attaining significance, including tests of haplotype sharing amongst affected pairs within and between families.